* Friday, 2011-11-25 at 12:38 -0600 - John E. Clifford <kali9putra@yahoo.com>:
> On Nov 24, 2011, at 9:44 PM, Martin Bays <mbays@sdf.org> wrote:
>
> > * Wednesday, 2011-11-23 at 13:34 -0800 - John E Clifford <kali9putra@yahoo.com>:
> >
> >> Summary
> >
> > Good idea!
> >
> >> 1. {zo'e}, as implicit in unfilled places, can't mean either "what I (would
> >> have) had in mind" or a particular quantifier, because there are too many cases
> >> where it has to mean the other.
> >
> > Pardon?
>
> What is obscure here?
It was just the English language being its usual annoyingly ambiguous
self - I couldn't disambiguate the relative scopes of "can", "not" and
"either". After your restatement, I'm interpreting it as
"not ((possible:...) or (possible:...))". I hope that is what was
intended!
> {zo'e} can't be either one of these because both of them occur as
> reasonable expansions into the blank space and being one would
> preclude being the other. Unless you mean that a particular
> quantifier is one thing I might have had in mind. But that creates
> the problem of mixing variables with names, which is not where we want
> to go, I think.
OK. So we need a meaning for {zo'e} which has each of these two as
special cases. This could be "particular quantifier over a domain
I (could) have in mind" - a special case being that it's quantification
over a singleton domain, so equivalent to it just being a constant
I (could) have in mind.
> >> It can't be {zi'o} either, since that really does both the sense
> >> and reference of the underlying predicate (think of all those
> >> places which can't be gone to from anywhere by any route on any
> >> means of transportation -- the center of the Earth, say, pace Edgar
> >> Rice Burroughs).
> >
> >> Ideally (I think), unfilled places should be particular quantifiers,
> >
> > Surely not just that. {ta melbi} is quite a different assertion from {ta
> > melbi da}.
>
> {ta melba} presumably means {ta melba mi}, so is a case where the fill
> should be made or { zo'e} used.
It could also be something "generic" - something like {ta melbi lo'e
prenu} (whatever that means).
There's the related thorny issue of observer places - although {sance}
is just "x1 is a sound emitted/produced by x2", so trees are no issue,
{carmi} is "x1 is intense/bright/saturated/brilliant in property (ka) x2
as received/measured by observer x3". Is a candle's light carmi (be fi
zo'e) when there's no-one around to see it? Or is it only carmi be fi
zi'o?
> (I don't expect this to happen except in really formal writing -- and
> probably not even there -- but we're talking about feed into semantics
> her and how to fill in gaps.)
> >
> >> {zo'e} should be stated when a fixed, though perhaps unspecified,
> >> referent is intended.
> >
> > I think having a word which literally acts as if the place were unfilled
> > is a useful enough feature that we shouldn't do away with it unless
> > necessary.
> >
> > Perhaps we can use {lo du} for the meaning you suggest?
>
> I think I am missing your point here. {zi'o} says the place is
> unfilled. {zo'e} says the place is filled but I'm but telling you by
> what. And what does {lo du} do? It is either the self-identical
> things, which provides no information,
Yes, that was the intention. So it would have the meaning you're
suggesting for {zo'e}, whether or not {zo'e} itself does.
(Except that it only works if the unfilled second place of {du} is
interpreted correctly... it would be nice to have a clearer way of
getting at the always-true unary predicate. Do we have one?)
> >> While I'm at it, we should change {ce'u} over to a variable-binding
> >> operator so we can do abstractions right.
> >
> > Pardon
>
> Make it be lambda and put variables after it, so we can distinguish
> when two arguments are the same from when they are different.
We can already get that effect by using anaphora - {lo ka ce'u ri broda}
is unary, while {lo ka ce'u ce'u broda} is binary.
> >> 2. {lo broda} refers to a bunch of brodas (either an L-set or a plural
> >> reference, as your ontological conscience guides you), fixed by context but
> >> possibly not terribly specific. The bunch may have a single member or encompass
> >> all brodas that have ever been and maybe more (all in this universe of
> >> discourse, of course, though maybe not in this world
> >
> > But all satisfying broda(_) in this world, right, whether or not
> > zasti(_)?
> >
> Yes, I suppose so.
>
> > (This relates to maikxlx's intensionality remarks)
> >
> >> ). These latter, maximal, bunches represent brodakind for all
> >> practical purposes. Because of the transparency of bunches, such
> >> a bunch of brodas is also a bunch of kinds of brodas, etc. These
> >> maximal bunches might usefully have a separate gadri.
> >> [...]
> >> 3. Bunches relate to predicates in a variety of ways, for none of which does
> >> Lojban have an explicit marker, though some can be inferred from other factors
> >> (quantifiers, modals -- though we are somewhat defective there as well, or maybe
> >> just more pragmatic or rhetorical devices -- I'm not sure what generalization or
> >> stereotype is). I don't have a complete list and am unsure about the status of
> >> some I do have, so some discussion would be welcome.
> >
> > Right, this is the part of your approach I'm unhappy with. I'm loath to
> > give up the simple version of plural semantics, whereby a selbri is
> > interpreted in a given world just as a relation on the set of bunches.
> >
> But as far as I can see, you are the one who has given that up.
> I certainly have not. I have nothing but bunches of broda all the way
> up (or down, as the case may be). I am deliberately avoiding the use
> of"set", since that raises other problem. So far as I am concerned,
> the domain of the functions can be just bunches (of bunches, if you
> like).
>
> > Complicating this with your "modes of predication" (conjunctive,
> > disjunctive, collective, statistical...) seems to fit lojban ill,
> > precisely because lojban has no way to mark them.
>
> So far as I can tell, this jumble just comes with plural reference,
> together with an attempt to realistically with how various things we
> say are related to the things we talk about. I don't suppose the list
> is complete yet, but that is only a practical problem. As for not
> fitting Lojban, Lojban was designed without plural reference (or
> L-sets) and so makes no allowance for them. What it does partly make
> allowance for is using (C-)sets to represent plurals. But just what
> that involved was never spelled out too clearly (and much of what was
> originally spelled out was lost in xorlo), so we cannot merely take it
> over for L-sets. Restoring some of that, or devising new conventions,
> can cover much of the difficulty and perhaps all, depending on how
> wide our convention net spreads.
Ah, so it looks like I have been misunderstanding you. I understood you
as having the truth value of a predication (in a world) depend on three
things - the predicate, the bunches which are its arguments, and the
mode(s) of predication. Now I'm understanding you as saying that it
depends only on the first two, with the mode(s) merely being a way of
describing how it is that the truth value is related to the truth values
of the various predications where the bunches are replaced by their
subbunches. Is that right?
Assuming it is - what makes me uneasy is the shift from one mode to
another when our bunches get large enough.
In {ro lo verba cu prami lo mamta}, if {lo mamta} is interpreted as
a few specific mothers, presumably prami acts distributively in x2
giving the meaning that each of the children love each of those mothers.
But if we enlarge the bunch to the maximal bunch of mothers, it switches
to be (something like) disjunctive - now we're just saying that each
child loves some mother (or maybe just Mother, if that's somehow
different?), perhaps their own, and not that each loves every mother.
So where did this switch occur on the way from our little bunch to the
maximal bunch, and why? Does it have something to do with adding mothers
which don't exist? Does adding just one non-existent mother cause the
switch?
> > The alternative is to further complicate the domain - adding more
> > derived entities beyond bunches. The marking can then be done with gadri
> > and quantifiers.
>
> I'm not sure what you mean here. I don't see how adding new entities
> (what, I wonder
Things like the kind Mother.
> ) will help with the modes of predication issue. A few
> nice adverbs seem to be the most natural way to proceed.
So this would be explicitly marking which mode of predication is meant
to be in use, hence giving joint information about the precise predicate
intended (when there's vagueness in that) and the bunches intended?
> >> 4. We need a way to sort out the official meaning (sense, a function on worlds)
> >> and the ordinary meaning, an area in in the web of other meanings (probably not
> >> a spot in the Platonic tetrahedron anymore). And then say which one we are
> >> talking about.
> >
> > Pardon?
>
> Well, it seems to me that people (myself included) flop back and forth
> among "lion" represents a function from(or relation between) things
> and truth values and "lion' represents a function from worlds to a set
> of objects in that world and "lion" represents the property of being
> a large cat (genus Panthera) ... . Trying to satisfy the various
> conditions these impose is a problem, since they are very different
> . I think part of our problem is that we often are at cross purposes
> here.
I see the first as being what you get from the second after specifying
a world, and the third as being another way of looking at the second.
Martin
Attachment:
pgpOLIiu33uOC.pgp
Description: PGP signature